The AtVAM3 Encodes a Syntaxin-related Molecule Implicated in the Vacuolar Assembly in Arabidopsis thaliana *

The vacuole constitutes a large compartment in plant and fungal cells. The VAM3 gene ofSaccharomyces cerevisiae encodes a syntaxin-related protein required for vacuolar assembly. An Arabidopsis thalianacDNA library, designed for expression in S. cerevisiae,was screened for cDNAs able to complement defective vacuolar assembly of the Δvam3 mutation. One cDNA, encoding a 33-kDa protein with structural similarities to the other syntaxins, was identified. The product of AtVAM3 (AtVam3p) was expressed in various tissues including roots, leaves, inflorescence stems, flower buds, and young siliques. The AtVAM3 transcripts were abundant in undifferentiated cells in the meristematic region. AtVam3p fractionated predominantly to an 8,000 × g pellet fraction where a vacuolar membrane protein H+-translocating inorganic pyrophosphatase (H+-PPase) also fractionated. Immunoelectron microscopy showed that AtVam3p was localized to restricted regions on the vacuolar membranes. We propose that AtVam3p provides the t-SNARE function in the vacuolar assembly in A. thaliana.

The vacuole constitutes a large compartment in plant and fungal cells. The VAM3 gene of Saccharomyces cerevisiae encodes a syntaxin-related protein required for vacuolar assembly. An Arabidopsis thaliana cDNA library, designed for expression in S. cerevisiae, was screened for cDNAs able to complement defective vacuolar assembly of the ⌬vam3 mutation. One cDNA, encoding a 33-kDa protein with structural similarities to the other syntaxins, was identified. The product of AtVAM3 (AtVam3p) was expressed in various tissues including roots, leaves, inflorescence stems, flower buds, and young siliques. The AtVAM3 transcripts were abundant in undifferentiated cells in the meristematic region. AtVam3p fractionated predominantly to an 8,000 ؋ g pellet fraction where a vacuolar membrane protein H ؉translocating inorganic pyrophosphatase (H ؉ -PPase) also fractionated. Immunoelectron microscopy showed that AtVam3p was localized to restricted regions on the vacuolar membranes. We propose that AtVam3p provides the t-SNARE function in the vacuolar assembly in A. thaliana.
The vacuole in plant and fungal cells constitutes a large compartment with various physiological functions (1,2). The vacuole serves as a storage compartment for various primary and secondary metabolites including amino acids, organic acids, and sugars. It is a lytic compartment as well, sharing common features with lysosomes of mammalian cells. It also functions in homeostatic regulation of cytosolic ions by transporting small molecules across the vacuolar membrane through various transporters. One of the most prominent features of the plant vacuole is its large volume. In fully matured plant cells, it occupies over 90% of cell volume. Individual plants need to expand their stems, leaves, and roots to acquire resources from the environment including light energy, minerals, and water. This expansion in plant body size, however, would be costly if it were accomplished by cell division or by synthesis of cytosolic material. In fact, cell enlargement is mostly attributable to the increase in vacuolar volume. Therefore, the plant vacuole provides a space-filling compartment (3).
While the physiological importance of the plant vacuole is clear and many of the molecules that make up this compartment have been well characterized, the molecular mechanisms involved in its assembly are largely unknown. In contrast, vacuolar biogenesis and assembly in the unicellular organism Saccharomyces cerevisiae are fairly well understood. Our previous genetic studies showed that nine VAM genes (for vacuolar morphology) are involved in the assembly of yeast vacuolar compartments (4). We have also shown that the yeast VAM3 gene encodes a member of the syntaxin family (5), the key molecules regulating vesicular transport (6,7). Syntaxins are membrane receptors on the target membrane (t-SNARE) that interact with the other molecules on the transport vesicles (v-SNARE). Specific interactions between the t-SNARE and v-SNARE molecules ensure fusion of the transport vesicles with their target membranes and the correct delivery of their contents (8). The yeast vam3 mutants show a characteristic phenotype; they accumulate numerous small vesicles instead of large vacuoles, and they are defective in the processing of vacuolar membrane and soluble proteins. We have also shown that the product of VAM3 (Vam3p) is localized to the yeast vacuolar membranes (5). These structural and functional features of Vam3p are consistent with Vam3p providing the t-SNARE function in the vacuolar assembly in yeast cells. Vam3p is the first t-SNARE identified on the vacuole/lysosome in eukaryotic organisms.
Toward understanding the development of vacuolar compartments in higher plant cells, in the present paper we searched for the vacuolar syntaxin of Arabidopsis thaliana. We predicted that vacuolar assembly in yeast and plant cells may share similar molecular machinery. If so, plant molecules may replace the function of corresponding yeast molecules, as shown in a handful of cases in different stages of intracellular vesicular trafficking (9 -12). We predicted that the vacuolar syntaxin of A. thaliana could substitute the Vam3p function in yeast cells. Here we report on the identification and characterization of a new member of syntaxin-related molecule, AtVam3p, the vacuolar t-SNARE of A. thaliana.

EXPERIMENTAL PROCEDURES
Yeast Strains-Yeast vam3 strain YW22-12B has been described previously (5). The PEP12 gene of S. cerevisiae (13) was obtained by PCR 1 amplification. A genomic DNA fragment containing PEP12 was screened for a yeast genomic library and subcloned into pBluescript II SK ϩ (Stratagene). A pep12::LEU2 fragment was constructed by replacing the SplI/PstI region of PEP12 with the PvuII-PstI fragment of pJJ282 (14) containing LEU2 and was used for disruption of the chromosomal PEP12 gene by the standard method (15). Yeast cells were cultured in SCD or SCGS supplemented appropriately (16,17).
Molecular Cloning of the AtVAM3 Gene-The A. thaliana cDNA library in YES (18), originally constructed by R. W. Davis of Stanford University, was provided from Y. Komeda of Hokkaido University. The library was introduced into the yeast ⌬vam3 strain by the modified lithium acetate method (19,20). Ura ϩ transformants were selected on SCGS (ϪUra) containing 10 g/ml adenine. Candidates for Vam ϩ colonies were identified by strong red pigmentation on the plates and subsequently verified by fluorescence microscopy for ade pigment (5,16,17,21). Plasmid was recovered and the cDNA insert was introduced into pBluescript II SK ϩ (Stratagene) and sequenced on both strands. The coding region of the AtVAM3 cDNA was amplified by PCR, verified by sequencing, and placed under the control of the TDH3 promoter in a plasmid pKT10 (2 m URA3) (22). The resulting plasmid was designated pKT10-AtVAM3.
DNA and RNA Gel Blot Analyses-Genomic DNA blot was hybridized with the 32 P-labeled, randomly primed full-length AtVAM3 cDNA fragment in a hybridization solution containing 6 ϫ SSC, 5 ϫ Denhardt's solution, 1% SDS, and 100 g/ml salmon sperm DNA at 60°C overnight. The membrane was then washed once with 2 ϫ SSC and 0.5% SDS at room temperature for 30 min, and twice with 1 ϫ SSC and 0.1% SDS at 60°C for 15 min.
The 3Ј-untranslated region of AtVAM3 was amplified by PCR using a pair of oligonucleotides, SAT1-NCRA (5Ј-AATGAGCTCATAGTA-CTCGCAGCT-3Ј) and SAT1-NCRS (5Ј-AATGGTACCCCACAAACA-AAACAGAGTTTCTGT-3Ј), and subcloned into pBluescript II SK ϩ to generate plasmid pMHS005. Total RNA was isolated by RNeasy Plant Total RNA Kit (QIAGEN). Thirty g of RNA was electrophoresed and transferred onto a nylon membrane and then hybridized with 32 Plabeled RNA probes produced from plasmid pMHS005. The blot was washed in the same way as the DNA blot.
RNA in Situ Hybridization-Seedlings (7 days old) were fixed in 4% paraformaldehyde and 0.25% glutaraldehyde in 0.1 M sodium phosphate buffer, pH 7.4, overnight at 4°C, dehydrated through a graded ethanol series followed by a xylene series, and embedded in Paraplast Plus (Sherwood Medical). Microtome sections (4 m) were deparaffinized in xylene, rehydrated through a graded ethanol series and then hybridized with digoxigenin-labeled RNA probes prepared from the full-length cDNA of AtVAM3 by T3 or T7 RNA polymerase as described previously (23).
Protein Blot Analyses-The cDNA fragment encoding a part of AtVam3p (Arg 30 -Lys 200 ) was introduced into pGEX5X-1 (Pharmacia Biotech Inc.), and a GST-AtVam3p fusion protein was produced in Escherichia coli NM522. The fusion protein was purified on a glutathione-Sepharose 4B column (Pharmacia) and used for immunizing a rabbit. Arabidopsis tissues were frozen in liquid N 2 , mixed with 1 ϫ SDS-polyacrylamide gel electrophoresis sample buffer, and homogenized in Eppendorf tubes with disposable pellet pestles (Kontes). The lysates were boiled and centrifuged to remove debris. Preparation of yeast cell lysates and immunoblotting analyses were described previously (16,17,24). Subcellular Fractionation of A. thaliana Cellular Membrane-A. thaliana was grown in liquid medium (0.5 ϫ Murashige and Skoog salts and vitamins (Gibco), 2% sucrose) at 20 to 25°C under 16 h light/8 h dark condition for 3 weeks. Roots were excised, frozen in liquid N 2 , and then ground to fine powder. The powder was resuspended in AT extraction buffer (0.1 M Tris-Cl, pH 7.5, 0.25 M sucrose, 1 tablet of complete proteinase inhibitor (Boehringer Mannheim) per 50 ml of buffer) and then centrifuged at 500 ϫ g for 15 min to remove debris and unbroken materials. The cell lysates were spun at 8,000 ϫ g for 15 min to give a pellet fraction (low speed pellet (LSP)) and then the supernatant was further spun at 100,000 ϫ g for 1 h to give pellet (high speed pellet (HSP)) and supernatant (high speed supernatant (HSS)) fractions.
Immunoelectron Microscopy-The apical shoot meristem of 1-week old A. thaliana was fixed for 2 h with a fixative that consisted of 4% paraformaldehyde, 1% glutaraldehyde, and 0.06 M sucrose in 50 mM sodium cacodylate, pH 7.4. After washing with 50 mM sodium cacodylate, pH 7.4, that contained 0.06 M sucrose, the specimen was dehydrated in a graded series of dimethylformamide at Ϫ20°C and embedded in LR-White resin (London Resin Co. Ltd., UK). ImmunoGold procedures were essentially the same as those described earlier (25), except for the use of antiserum against GST-AtVam3p (diluted 50-fold) and 30-fold diluted protein A-gold (15 nm; Amersham Corp.). The sections were examined with a transmission microscope (model 1200EX; JEOL).

RESULTS
Identification of VAM3-related Gene in A. thaliana-Yeast vam3 mutants lack normal large vacuoles but accumulate small vacuole-related structures. This Vam phenotype is clearly seen even on plates by colony pigmentation of ade fluorochrome (4). This pigment is produced in the ade1 and/or ade2 mutants and accumulates in vacuoles (26). Vam ϩ ade2 cells develop red colonies while vam3 ade2 mutants form less pigmented pink colonies, presumably due to the abnormal vacuolar morphology and reduced vacuolar volumes in the vam3 mutant cells. To obtain the VAM3-related genes from A. thaliana, we screened cDNAs that were able to complement the defective vacuolar morphology of the vam3 mutant. We obtained 49 red candidates among 1 ϫ 10 5 transformants and observed them under a microscope to verify vacuolar morphology. Among them, we found one clone that developed large prominent vacuoles (Fig. 1). The plasmid was isolated from this Vam ϩ transformant, and it was reintroduced into the ⌬vam3 cells to confirm that the complementing activity was linked to the plasmid. The library expresses the cDNAs under the regulation of the yeast GAL1 promoter that is highly induced in the presence of galactose as a carbon source. The transformants never developed large vacuoles in the glucose medium (Fig. 1A) where the expression from the GAL1 promoter was repressed. In contrast, their vacuoles acquired normal morphology in the presence of galactose (Fig. 1B).
We determined the nucleotide sequence of the cDNA and found an open reading frame encoding a 29.5-kDa polypeptide of 268 amino acid residues ( Fig. 2A). The size of the isolated cDNA (1183 base pairs) agreed well with the size of its RNA transcript (1.2-kilobase, Fig. 5B), thus we concluded that this cDNA represents the full-length transcript. The amino acid sequence shows structural similarities to the members of syntaxin, including yeast Vam3p (Fig. 3). We designated this Arabidopsis gene, AtVAM3, from its structural similarity and functional potential for complementing the yeast vam3 mutation. The product, AtVam3p, has a heptad repeat structure (residues 164 -220) in which hydrophobic amino acid residues appear at seven amino acid intervals. Such regions have a high potential to form an amphiphilic ␣-helix, intriguing for the intermolecular interactions by forming coiled-coil structure. AtVam3p has a highly hydrophobic segment at its C terminus thus implicating it to be a membrane protein while the rest of the sequence is hydrophilic, and there is no typical signal sequence for membrane insertion at the N terminus (Fig. 2B). The other members of the syntaxin-related proteins such as Pep12p (13) characteristics. They are hydrophilic proteins with 280 -330 amino acid residues, and they possess single hydrophobic transmembrane domains at the C terminus but lack the typical signal sequences for membrane translocation at the N terminus. The amino acid sequence of AtVam3p was aligned to the syntaxin-related molecules (Fig. 3). AtVam3p shares 27.2 and 21.6% identity to Pep12p and Vam3p over entire length, respectively. Interestingly, AtVam3p failed to replace the Pep12p function in yeast cells (see below) although the structural similarity predicts that AtVam3p may relate more closely to yeast Pep12p than to Vam3p.
Functionality of AtVam3p in S. cerevisiae-The PEP12 gene of yeast encodes another syntaxin-related molecule that is also involved in transport between the Golgi and vacuoles in yeast cells. Pep12p and Vam3p are both syntaxin-related molecules and involved in the vacuolar biogenesis; however, these two proteins provide the t-SNARE function at different stages of yeast vacuolar assembly. Pep12p is shown to be localized on the prevacuolar/endosomal compartments (13), whereas Vam3p is found in the vacuolar membranes (5). Deletion of their function results in distinctive phenotypes. ⌬pep12 cells exhibit mostly single large vacuoles, a characteristic phenotype of the class D vps mutant group (32) to which pep12/vps6 belongs, while ⌬vam3 cells accumulate multiple small fragmented vacuoles.
⌬pep12 cells have severe defects in delivery of a set of vacuolar proteins including carboxypeptidase Y (CPY) to the vacuole. The pep12 mutants mistarget a considerable amount of newly synthesized CPY to the cell surface, thus ⌬pep12 cells contain essentially no mature CPY (13). In contrast, ⌬vam3 cells do not secrete CPY, though processing of CPY is less efficient in ⌬vam3 than in the wild-type cells. Therefore, ⌬vam3 cells contain both a mature form CPY as well as abnormally processed CPY (5).
We found that the vacuolar morphology of yeast ⌬pep12 cells did not change upon introduction of the AtVAM3 gene (data not shown), indicating that the AtVAM3 could not complement the pep12 mutation. The differential defect in the maturation of CPY provided a much clearer clue for assessing the functionality of AtVam3p in yeast cells. When the AtVAM3 was expressed from a constitutive promoter of TDH3, the amount of mature CPY (mCPY, 61-kDa form) increased, showing that the expression of the AtVAM3 complemented the defective maturation of CPY in ⌬vam3 cells (Fig. 4). However, the defective maturation of CPY was not complemented by TDH3p-AtVAM3 in the ⌬pep12 mutant, as ⌬pep12 cells harboring the TDH3p-AtVAM3 construct accumulate only a limited amount of Golgimodified p2-CPY (69-kDa form) but lacked mCPY, like ⌬pep12 cells harboring a control plasmid. This observation showed that the AtVAM3 failed to replace the function of PEP12 in yeast cells. In summary, AtVAM3 complemented the defective vacuolar morphology (Fig. 1) and CPY processing (Fig. 4) in the ⌬vam3 cells while it did not provide enough activity for rescue of the defective vacuolar biogenesis in the ⌬pep12 mutant.
Characterization of the AtVAM3 Gene in A. thaliana-DNA blot analysis probed with the full-length AtVAM3 cDNA showed multiple bands when the blot was washed under low stringency conditions, suggesting that the AtVAM3-related gene exists in the A. thaliana genome (Fig. 5A). The coding regions of the AtVAM3 and aPEP12 share approximately 70% identity at the nucleotide sequence level, and the sizes of the weakly hybridizing bands agreed with the reported restriction patterns of the aPEP12 gene though the ecotypes used in these studies were different. Here we used ecotype Columbia while the aPEP12 gene was characterized in ecotype RLD (27).
The spatial expression pattern of AtVAM3 in plants was examined by Northern hybridization (Fig. 5B). A transcript of 1.2 kilobases was identified, which agreed well with the length of the cDNA insert in the original clone. The expression of the AtVAM3 was seen in all tissues but at varying levels. In stems and roots, it was highly expressed, whereas the amount of transcript in leaves was relatively low, about one third of the level seen in stems and roots. We focused on the expression pattern of AtVAM3 around a shoot apical meristem of young plants by in situ hybridization analysis (Fig. 6). High uniform levels of AtVAM3 transcript were detected in morphologically undifferentiated cells. The signal was much higher in these cells than more mature, vacuolated cells. High levels of expression were also detected in the leaf primordia and vascular tissues.
Characterization of the Product of the AtVAM3-To facilitate analyses of the protein product, we prepared a rabbit antibody for a GST-AtVam3p (residues 21-200) fusion protein. The specificity of the antibody was verified by the yeast expression system (Fig. 7A). The antibody recognized a 33-kDa protein in yeast cell lysates prepared from the galactose grown ⌬vam3 cells harboring the GAL1p-AtVAM3 construct. This protein was not found when the yeast was grown in the glucose medium, an indication that the 33-kDa protein is the product of AtVAM3. The antibody recognized a protein in the A. thaliana lysate which shows the same migration in an SDS-polyacryl- amide gel as AtVam3p produced in yeast cells (Fig. 7B). This protein was not recognized by the preimmune serum (data not shown). From these observations, we concluded that the 33-kDa protein is the product of the AtVAM3 gene.
A similar expression level of AtVam3p was observed in var-ious tissues (roots, leaves, stems, flowers, and green siliques) when samples were normalized for overall protein content (Fig.  7B). We noted that the amount of AtVam3p in leaf was comparable with that in roots and stems while mRNA levels were lower in leaves than in roots and stems (Fig. 5). The subcellular localization of AtVam3p was examined both by subcellular fractionation by differential centrifugation and immunoelectron microscopy (Fig. 8). We speculated that AtVam3p may be localized on the vacuolar membranes because of its structural and functional similarities to the yeast vacuolar syntaxin, Vam3p. A post-nuclear supernatant of root was fractionated into LSP, HSP, and HSS fractions. Approximately 80% of AtVam3p was fractionated into the LSP fraction, and the remainder was found in the HSP fraction. A vacuolar membrane marker protein, H ϩ -translocating inorganic pyrophosphatase (H ϩ -PPase) showed a similar distribution in the differential centrifugation fractionation. This distribution pattern of AtVam3p supported our prediction that AtVam3p is localized to the vacuolar membranes in plant cells.
Ultrathin sections of shoot apical meristem of A. thaliana seedlings were probed with the antibody against AtVam3p and protein A-conjugated colloidal gold. The gold particles were found on the membranes of small vacuoles (Fig. 9, A and B). In a mature cell with a larger vacuole, the staining was more dispersed (Fig. 9C). This observation was consistent with the FIG . 3. Alignment of AtVam3p, aPep12p from A.thaliana, Vam3p, and Pep12p from S. cerevisiae. Alignments were generated by the MegAlign program from the LASERGENE package (DNASTAR, Inc.). Amino acid residues identical to AtVam3p are highlighted.

FIG. 4. Functionality of AtVam3p in yeast syntaxin mutants.
Total cell lysates from ⌬vam3 and ⌬pep12 cells harboring vector (pKT10) or TDH3p-AtVAM3 (AtVAM3) constructs were analyzed by immunoblotting using anti-CPY antibody and alkaline phosphataseconjugated secondary antibody. The endoplasmic reticulum, Golgi, and mature forms of CPY were indicated as p1-CPY, p2-CPY, and mCPY, respectively. The position of the 69-kDa molecular mass marker is indicated. , and hybridized with a cDNA AtVAM3 probe. The size marker is HindIII-digested -DNA and is shown in kilobase pairs. B, total RNA (30 g) from leaves, stems, and roots of A. thaliana was separated in a 1.2% agarose-formaldehyde gel, and the RNA blot was hybridized with a 32 P-labeled RNA probe corresponding to the 3Ј-untranslated region of AtVAM3. relatively lower transcript levels in the mature cells containing larger vacuoles (Fig. 6). AtVam3p was not distributed homogeneously on the vacuolar membrane. Rather, it appeared to accumulate in vacuoles that were interacting with one another (Fig. 9, A and C, arrows). This suggests that AtVam3p functions in membrane-membrane interactions. A higher concentration of preimmune serum did not give any signal (data not shown). We concluded from both subcellular fractionation and immunoelectron microscopic studies that AtVam3p is localized on the vacuolar membrane of plant cells. DISCUSSION Syntaxins are the key molecules regulating intracellular vesicular traffic. The primary structure of syntaxins, however, are not highly conserved enough to allow design of PCR primers or hybridization probes. We used a functional screening strategy to find a VAM3 homologue from A. thaliana and iden- tified a novel syntaxin-related molecule, AtVam3p. AtVam3p is structurally related to the yeast vacuolar syntaxin, Vam3p, it is able to replace the Vam3p function in yeast, and it is localized to the vacuolar membranes in plant cells. Taken together, we propose that AtVam3p provides the t-SNARE function at a certain step of the vacuolar assembly in A. thaliana.
Bassham et al. (27) identified another syntaxin-related protein in Arabidopsis, aPep12p, by a similar approach using the yeast pep12 mutant. Our studies together endorse the great potential that heterologous expression and functional screenings have, even in distantly related organisms such as A. thaliana and S. cerevisiae. AtVam3p and aPep12p appear to be closely related in their structures (over 60% identity at the amino acid level). This argues that these two syntaxins represent redundancy in the A. thaliana genome. However, we do not believe they are executing the same function. In yeast, we and others have shown that pep12 mutants and vam3 mutants have different phenotypes, clearly indicating that they are involved in different steps of the vacuolar assembly in yeast cells (4,5,13). The expression of the AtVAM3 did not complement the defective vacuolar targeting and maturation of CPY in a pep12 mutant (this study), while aPep12p provides the Pep12p function in yeast cells (27). These observations predict that AtVam3p executes its function in a similar step where Vam3p does in yeast, while aPep12p performs its role in accepting the Golgi-derived transport vesicles on the prevacuolar/ endosomal compartments as Pep12p does in yeast (13).
Since these studies addressed the functionality in yeast cells, we cannot rule out the possibility that AtVam3p and aPep12p provide redundant functions at a same stage of the vacuolar assembly in plant cells. We should also point out that heterologous expression and subsequent complementation alone may reflect functional cross-suppression or an artifact due to overexpression and mislocalization of the protein in question, especially when its expression is driven by a strong promoter like the GAL1 or TDH3 promoter. Therefore, the precise function of AtVam3p and aPep12p must ultimately be examined in plant cells. We have shown here that AtVam3p is localized to the vacuolar membranes. Most recently, aPep12p was reported to be localized to post-Golgi, endosome-like compartments that are clearly distinct from the vacuole (33). This differential subcellular localization of these two related proteins supports our view that aPep12p and AtVam3p perform distinct functions at different stages of the vacuolar assembly.
The vacuolar compartments in plant cells show heterogeneity; at the least, storage vacuoles and lytic vacuoles are distinct compartments containing different sets of proteins in their lumen and membranes (34). In addition, these two compartments undergo dynamic exchange and fusion during the plant growth and development (34 -36). These dynamic processes of the plant vacuoles are also suggested by the studies of vacuolar protein targeting signals for various vacuolar proteins. In plant vacuolar proteins, at least three different classes of protein targeting signals are involved in protein sorting (37)(38)(39)(40), and these signals seem to utilize different targeting mechanisms in the transport process from the Golgi to the vacuoles (41). As plant vacuoles are more diverse than their yeast counterparts both morphologically and biochemically, their assembly and dynamism must be regulated in a more sophisticated manner than in yeast cells. It is therefore likely that plant cells develop more elaborate systems for vesicular trafficking around the vacuolar compartments. According to the SNARE-hypothesis (6), each compartment should be distinguished by a specific tag, otherwise the specificity of the compartments would be compromised by nonspecific fusion of transport vesicles. We have shown here that AtVam3p was not uniformly distributed on the vacuolar membranes. Interestingly, yeast Vam3p exhibits a patchy localization to restricted regions of the vacuolar membrane (5) as is also true of the other yeast vacuolar assembly molecule, Vam6p (16). Large accumulations of AtVam3p on the membranes of small vacuoles suggests that AtVam3p is involved in vacuolar dynamics, for example, when two distinct, vegetative and storage vacuoles merge to develop large prominent vacuoles. These possibilities can be examined further by histological and immunocytochemical localization of AtVam3p during the plant developments.